The Coanda effect28 June 2004
Retrofitting Coanda screens to hydro power scheme intakes has made some sites more economically viable, says Matthew Palmer
COANDA SCREENS have now been installed at the intakes of more than 40 medium and high head hydroelectric schemes in Europe. These screens are static and self-cleaning, requiring no power source and negligible maintenance. Most of the installations have been at new-build sites, but several recent installations have been retrofitted at sites where the existing intakes have been problematic. This paper outlines the problems, engineering solutions and operational improvements of three of these installations. In all three cases, plant operation has improved dramatically. Two of the sites have demonstrated a payback period of only two years while the third may have been decommissioned if Coanda screens had not been installed.
There are many different types of intake for hydroelectric schemes, with the final design for a particular site based on a combination of local geography and hydrology, engineering experience and economics. For medium and high head schemes, traditional intakes vary from drop bar screens and Tyrolean riverbed intakes through to side intakes and channels with coarse screens and settling chambers. These all have advantages and disadvantages and have been well documented over the years. Unfortunately, intake design is often undervalued, even though poor design can lead to a significant long-term loss of generation and increased maintenance costs.
Coanda screens were developed in the US more than 20 years ago (see www.aquadynescreen.com). Initially designed for irrigation and fisheries, they have since been used highly successfully for hydro scheme intakes. They have no moving parts and are self-cleaning under most conditions, meaning that they require no power supply and negligible maintenance; these factors have very positive economic benefits.
The basic principle of Coanda screen operation is shown in Figure 1. The screen is installed on the downstream side of a weir, and water flows over the curved acceleration plate and down the screen, which is made from tilted wedge wire with a typical aperture of 1mm. The high capacity is a result of the shearing effect produced by the tilted wires, combined with the Coanda effect (fluid following a surface), which pulls water around the bars and through the screen. The geometry of the screen determines the capacity (Wahl, 2001) as well as the self-cleaning ability of the screen – flatter screens have higher capacity but tend to build up more debris.
More than 40 Coanda screens have been installed at European hydro sites over the last 10 years, mainly in the Alps and the UK (for list see www.dulas.org.uk). Whereas most of these sites have been new-build designs incorporating the Coanda screens, several recent installations have involved retrofitting the screens on existing intake structures.
The first constraint with Coanda screens is the head loss between the weir crest and the foot of the screen; typically 500-1300mm. This can be allowed for if the intake is being built specifically for such screens but can present difficulties if an existing intake is to be refurbished, especially if the hydraulic design is marginal. It is also the reason why such screens are not suitable for low head sites.
The second constraint is the length of weir required to produce a given capacity. Typical flow rates are 140l/sec per metre of weir, thus a flow rate of 1m3/sec would require a weir just over 7m long. If the structure has an existing weir, then it may be relatively easy to increase the crest height and build a sump chamber downstream, but increasing the width can often involve removing wing walls and possibly widening the river itself. In narrow gorges with bedrock this may prove impractical unless the weir can be extended along the line of the river.
If the crest cannot be raised – due to low upstream bank levels for instance – there may also be problems with air entrainment in the existing pipeline, depending upon the original hydraulic design. If the pipe remains at the same level, the mouth of the pipe could lose up to 1300mm of submergence. This problem can often be mitigated by ensuring laminar flow conditions to the pipe entry and by installing simple anti-vortex structures. Re-laying a section of low-gradient pipe to maintain positive pipe pressures is also possible, although potentially more costly.
In terms of environmental constraints, the main issue is the actual construction work itself. In the UK, as elsewhere, there are strict controls over carrying out concrete works in water courses. Once installed, however, environmental regulating bodies often prefer Coanda screens to standard bar screens, as the typical 1mm spacing prevents even the smallest fish and fry from entering the pipeline.
In spite of the above constraints, the significant increases in generation and annual income due to retrofitting Coanda screens can make even major civil works viable, as illustrated by some of the following case studies.
Case study 1 – Lodore Falls, England
A Coanda screen was installed at the Lodore Falls hydro scheme in the UK in 1999. This is a 170kW grid connected scheme, with a Turgo turbine running under a 150m gross head and a flow rate of 235l/sec. The Coanda replaced a wire basket screen, which was very difficult to keep clean. There had been particular problems during the autumn as the catchment is heavily wooded with broadleaf trees; at these times the basket screen would block within a few hours.
The installation itself was relatively straightforward as the existing weir was both wide and high enough to accommodate the new screen. This resulted in a relatively low cost installation, with civil works limited to building the sump chamber, refurbishing the weir crest and lowering the pipe slightly. Compensation flow was maintained by using a trough under the screen to divert part of the abstracted flow straight back into the river.
The Coanda screen at this site was the subject of a 15-month evaluation study by the Energy Technology Support Unit (ETSU, 2001), during which time the scheme operator estimated that there was a 15% increase in annual production from 700-800MWh, even allowing for the reduction in head due to the Coanda screen (about 1%). This resulted in an annual increase in revenue of over US$7000 which, combined with a reduction in the manual cleaning cost to almost zero, gave the screen installation a two-year payback.
Case study 2 – Iwrch, Wales
The hydro scheme at Iwrch in mid Wales is a 285kW grid connected Francis machine, operating under a 72m gross head with a design flow of 600l/sec. It was built in 2000, but inappropriate intake design resulted in very poor operational performance, to the extent that the plant ran briefly in 2002 before being shut down. In conjunction with a group of private investors, Dulas Ltd. subsequently bought the scheme and immediately redesigned the intake to allow installation of Coanda screens.
The original intake design consisted of a 5m long weir and side intake chamber with a vertical wedge wire screen (6000 x 430mm with 3mm aperture). The turbine was automatically controlled to keep the level in the intake chamber constant, so as the screens blocked and the flow into the chamber was reduced, the turbine gradually closed down to match what it saw as a reduction in river flow. The software had been modified so that at low flow rates the turbine would shut down, creating zero velocity through the screens and allowing leaves that had been held onto the screen to be washed away. The plant would then wait a short time before restarting.
During the period shown in Figure 2 this cycle lasted about 8hr, with the plant shutting down when the flow was around 150l/sec. It can be seen that the maximum flow was reduced slightly with each cycle, as the number of leaves that were not washed away increased. By estimating the area under the graph, the energy output for this period was only 56% of that available. It should be noted that Figure 2 refers to February, when leaf load in the river was not particularly high. If the plant had been running during the autumn, the cycling time would be greatly reduced, with even greater loss of generation.
The high bed load of gravel also caused problems for this intake design. During high flood flows the weir pool would become completely filled, thus blocking the screens. The operator would then have to wait until the river flow was low enough to enable a digger to enter the river and remove the gravel. This meant that the turbine was often inoperable during periods of high flow availability.
In 2003 the intake was redesigned to incorporate Coanda screens, with the objective that the associated reduction in downtime would lead to a long-term increase in revenue. There were several issues that had to be overcome in order for the design to be feasible.
Firstly the weir needed to be extended from 5-8m in order to provide sufficient capacity through the Coanda screens. This required the removal of a mortared stone and concrete wing wall, which was easily achieved with a machine that made full use of the good access to this bank of the river. The weir extension also allowed oversized compensation flow pipes to be embedded in the structure, allowing the river flow to be diverted during the remainder of the work. Once completed, orifice plates were used to restrict the compensation flow to that agreed with the UK Environment Agency.
In order to connect the new sump to the pipeline, an intermediate chamber was built to allow water to flow back under the weir and into the existing bellmouth chamber. By incorporating as much of the original structure as possible, the design minimised the civil works required. However, the new arrangement created a non-ideal flow path (eddy formation) for the water entering the intake bellmouth. Coupled with the headloss of the Coanda screen, which reduced the submergence of the existing intake bellmouth, this created the potential for problems with air entrainment. This was mitigated by the following actions:
• Half height Coanda screens were used, which minimised the headloss to 700mm.
• The weir crest was raised by 200mm. This did not lead to an increase in upstream flood levels due to the extra weir width created to accommodate the Coanda screens.
• A floating grid was installed above the bellmouth intake. This prevented eddies from developing into vortices, hence avoiding any air entrainment.
The scheme was re-commissioned in December 2003, and to date there has been no downtime or reduction in output due to screen performance. Gravel is no longer a problem; the pool behind the weir has mostly filled with gravel, with any further gravel spilling over the weir and screens and continuing down the river. Hence there is no longer any requirement to visit the intake to clean screens or excavate gravel, with associated reductions in labour costs. Figure 3 shows a four-day period of operation during March 2004, where flow remains constant until the turbine begins to river follow. The improvement compared to the old intake is significant.
From hydrological models, the scheme is estimated to produce around 1100MWh per year. The plant only operated for a few months with the original screens, so no annual figures are available. However, the available data shows that poor screen operation caused output to be just over half of what was possible. This would have resulted in a loss of some 484MWh per year, worth US$44,000 at the current selling price of US$91/MWh. The total cost of the intake works was in the region of US$85,000 (including US$16,700 for the screens) giving a payback of about two years excluding the saving in labour costs for gravel removal and screen cleaning. During the first four months of 2004 the scheme generated 451MWh.
Case study 3 – Samina, Austria
The Wasserfassung Samina Werk 1 hydro scheme on the Samina river in Austria is operated by E-Werke Frastanz . Water passes from the main intake through a tunnel to a secondary screening building, from where a pipeline feeds two Francis machines (85m net head with flow rates of 1200l/sec and 1600l/sec). The outfall from these then powers a further two Ossberger crossflow machines. Total rated output for the four machines is just over 2.7MW.
The original intake structure was built about 50 years ago, and consisted of a drop bar screen, which then fed a small settling chamber before discharging into the tunnel. The screen was almost horizontal and therefore often blocked with stones, and the 50mm spacing meant that large quantities of gravel entered the system. Leaves and small sticks also passed all the way through to the turbines. The result was that the settling chamber needed to be flushed out approximately 200 times per year, and up to 600 times in years with heavy flooding. Each flushing procedure lasted 30min and typically lost 500kWh in generation – between 100-300MWh per year. The tunnel was also filling with gravel, to the extent that it had lost about half of its cross sectional area. The secondary screening building at the far end of the tunnel consisted of a bar screen with 20mm spacing and an automatic trashrack cleaning machine.
During 1999 and 2000 the existing intake was damaged in heavy flooding, and the whole structure therefore needed rebuilding. Without rebuilding the intake, generation would most likely have ceased completely at this site. It was initially thought that the only solution would be a large settling chamber excavated out of the mountain upstream of the intake, but cost estimates of US$910,000 for the chamber alone were prohibitive. However, by incorporating Coanda screens in the design, the rebuilding became economically viable.
Construction was completed in 2003. The drop bar screen was installed at a steeper angle, and the Coanda screens were placed in two rows on top of the old settling chambers. The linear length of the screens is 19.5m, with a rated capacity of 2660l/sec. The Coanda screens were not used as a replacement for the bar screen due to the threat of damage from excessive bed load during flood conditions.
With the new design, plant operation has been significantly improved. The steeper bar screen has less blockage problems than before. The channel between the Coanda screens still requires periodic flushing, but not as often, and the time taken to flush has been halved, thus reducing downtime and maintenance costs. Sand, gravel, leaves and sticks are prevented from entering the tunnel (the 1mm spacing of the screens excludes all particles over 1mm and most particles over 0.5mm) and the secondary trashrack and automatic cleaner are now redundant.
In the end, the total cost of the refurbishment work was about US$1.2M including a new access track and construction of a temporary intake – the screens themselves cost in the region of US$60,000. Bad weather and flooding also caused problems during construction. The operators have found it difficult to estimate the direct payback time of the refurbishment, but without installing Coanda screens the scheme would most likely have been decommissioned.
All refurbishment and retrofitting work is constrained by site geography and the geometry of existing structures, but with innovative design and engineering many problems can be overcome. Coanda screens have their own constraints – mainly head loss and weir width. The benefits in terms of increased output are very site specific, with the economics of retrofitting highly dependent on the design and performance of the original intake structure, and the requirements of the turbine. It can never be stated that Coanda screens should be retrofitted to all high and medium head hydro sites, since many traditional intakes operate satisfactorily, and have done for many years.
However, in the cases outlined above, the replacement of poorly designed intakes with Coanda screens has greatly improved output, to the extent that the increase in generation and reduction in maintenance cost far outweighs the capital cost of the new screens and the associated civil works, and the head lost by the screens themselves. Indeed in all three examples it could be argued that Coanda screens are responsible for making the sites economically viable.